Current latest usage ratings for various AI GPT score of 9, think mode is very easy to use, fast, accurate, and new. Claude score 8.5, possibly due to insufficient computing power, laziness, or mistakes Manus 7.5 points optimization is good, but too expensive Kimi 7-point multi-agent cluster is indeed good, but there are many errors Dou Bao has a weak ability of 6.5 points, but taking care of the emotions of the elderly and children Gemini scored 6 points very lazy, possibly learning to follow the bean bun route GROK wins by 6 points in X search, but with more computing power, it is estimated that @ elonmusk will exert more effort later Qianwen 6 Deepseek 5 points Mixed Yuan 2 points ERNIE Bot 1 point If you can only choose one, I would recommend GPT. If there is no risk of Claude being banned, he can also be the main player - if either side updates at any time, it will surpass the other and score very close.